Articles and reviews in leading tech publications

Anthropic Unveils Upgraded Claude AI Models with Enhanced Coding Skills and 'Computer Use' Feature

Oct 22, 2024, 03:12 PM

Anthropic has announced the release of an upgraded version of its Claude 3.5 Sonnet AI model and introduced a new model, Claude 3.5 Haiku, which matches the performance of Claude 3 Opus while offering better coding skills and high accuracy. The new Sonnet model features significant improvements in coding capabilities, scoring 49% on the SWE-bench Verified test, up from 33.4%, surpassing all publicly available models on coding tasks. Additionally, Anthropic introduced a new experimental 'computer use' feature, now in public beta, that allows Claude to interact with desktop applications as humans do—by analyzing what's on the screen, moving the cursor, clicking buttons, and typing text. This capability enables the AI to click, type, and browse the web on the user's behalf, opening up new possibilities for automated debugging, customer support, and educational applications. The updated models and the computer use feature are available for developers via API access through Anthropic, Amazon Bedrock, and Google Cloud Vertex AI.